Skip to main content

AI Usage > Lines added

Written by Tom Williams

Dataset: AI Usage

Entity: AI Usage Measurement

Field ID: lines_added

Type: Number

Description: The number of lines of code written by the AI assistant that reached the file. Compare with lines_suggested_added, which is what was merely offered.

Source: App

Transformation logic: N/A

App Mapping

GitHub Copilot

loc_added_sum

Anthropic Claude (coming soon)

core_metrics.lines_of_code.added

Cursor (coming soon)

acceptedLinesAdded

OpenAI Platform (coming soon)

N/A

OpenAI Codex (coming soon)

N/A

Reporting Use Cases

The Lines added field is AI output that actually landed in the codebase. Of all the measures in this dataset, it is the closest to a direct statement of AI contribution.

  • AI share of code written: Compare it against total additions in the Commits from PRs dataset over the same period to estimate the share of code written with AI assistance.

  • Contribution by person or team: Group by actor_username over aggregation_level = USER records.

  • Not a productivity measure on its own: Lines of code reward verbosity. Pair it with pull request and review measures rather than reporting it alone.

  • Three providers report it: The two OpenAI integrations publish no line counts.

Did this answer your question?